refresh rate USB

[David McFarlane]

Tobias,

Sorry, I do not have citations handy, I will leave that as an 
academic exercise for you.  But I believe this question, and a 
related one, has been covered over the years in publications such as 
"Behavior Research Methods", as well as in a white paper that I 
believe came from PST (and may appear in the E-Prime User Guide).

First, the rate-limiting issue seems to *not* be the transfer rate of 
USB, but rather the characteristics of the keyboard hardware.  Even 
if USB transfers at 120 Hz or higher, that does *not* mean that the 
device (in this case the keyboard) itself collects & sends data at 
that rate.  With keyboards especially, this seems to be a bad 
assumption.  First, individual keyboards are all over the map with 
regard to response latencies and variance, so even if you measure one 
you have no idea how that applies to others, even others of the same 
manufacturer and model!  With that warning, it seems that keyboards 
often have variable latencies in the ballpark of 0 - 15 ms.  This 
means that if you get an individual RT of, say 300 ms, you don't know 
if it was really 300 ms, or 285 ms, or somewhere in between.

So if you really need to know *individual* RTs down to the ms, then 
you need to use something other than a keyboard (e.g., SRBox; buttons 
on computer mice have been shown to have long but low-variance 
latencies, which could help, but mouse movement may interfere with 
this -- once again, search the literature, or better do your own tests).

But most studies either do not need that much precision, or do not 
need to know *individual* RTs.  In particular, the behaviors that we 
wish to measure often have variance that far exceeds the variance in 
our sloppy input devices.  And this was the thesis of the white paper 
that I referred to at the beginning.  The authors did some 
computations to show that, in almost every case, simply adding a few 
more trials or subjects to measure an average RT brings the variance 
down as good as we could get by going to great lengths to use a more 
precise input device.  So we should just quit fretting about this in 
most cases, and save that worrry for the few times that it really applies.

I will be interested to hear others weigh in.

-- David McFarlane, Professional Faultfinder
"You got to test that device performance, You got to test it for yourself,
No one else can test it for you, You got to test it for yourself."
(Apologies to the Fairfield Four)


>I plan an experiment in which Reaction Times for two conditions will
>be compared and not too much of a difference is expected. Let's say 15
>ms. I am wondering if an USB keyboard is the appropriate choice for
>such an experiment. I've heard that USB usually has a refresh rate of
>120 Hz, meaning that responses are checked every 8 ms. That's a lot of
>noise added to the real reaction times.
>
>On the other hand, EEG data and other stuff CAN be transferred much
>faster than 120 Hz via USB cable. THe question is, whether Windows (or
>E-Prime) can handle this.
>
>What is your experience here?
>
>Thanks a lot and so long,
>Tobias
>
>--
>You received this message because you are subscribed to the Google 
>Groups "E-Prime" group.
>To post to this group, send email to e-prime at googlegroups.com.
>To unsubscribe from this group, send email to 
>e-prime+unsubscribe at googlegroups.com.
>For more options, visit this group at 
>http://groups.google.com/group/e-prime?hl=en.

-- 
You received this message because you are subscribed to the Google Groups "E-Prime" group.
To post to this group, send email to e-prime at googlegroups.com.
To unsubscribe from this group, send email to e-prime+unsubscribe at googlegroups.com.
For more options, visit this group at http://groups.google.com/group/e-prime?hl=en.